Samsung Galaxy S10e =?utf-8?Q?can=E2=80=99t_?=reach local devices in network

[David Wibergh]

The command output is the same:


Sony

$ ip route get 10.220.0.4
10.220.0.4 dev wlan0 table 1026 src 10.220.0.6 uid 10356
    cache


Galaxy S10e

$ ip route get 10.220.0.4
10.220.0.4 dev wlan0 table 1017 src 10.220.0.5 uid 10294
    cache


We’ll try to get some of our users to run similar tests. We do have an S7 that we can dig up and do some testing on. If you want to go down the rabbit hole, but don’t have access to a galaxy device, I could ship the S10e to you.

Thanks,
David


På 9 oktober 2020 kl. 16:03:02, Jason A. Donenfeld (jason at zx2c4.com) skrev:
> Hmm, from that command output, I'm still not quite sure. The most
> significant difference I can see from casual inspection is:
>  
> Sony:
> 19000: from all fwmark 0xcd/0x1ffff iif lo lookup 1026
> 21000: from all fwmark 0xce/0x1ffff lookup 1026
>  
> Galaxy:
> 19000: from all fwmark 0x2ad/0x1ffff iif lo lookup 1017
> 21000: from all fwmark 0x2ab/0xffff lookup 1017
>  
> Is the galaxy missing a leading 0x1 in the mask of that second rule there?
>  
> What do you get on each when you run `ip route get 10.220.0.4`? Same
> thing or is it different?
>  
> I also noticed that on the Galaxy, there's a direct route to some
> telco server -- `90.132.128.59 via 10.220.0.1 dev wlan0 table 99 proto
> static`. I didn't see this direct route on the Sony. I doubt it's
> related to the above, but I've also never seen this before and wonder
> if it's some VoLTE/VoWIFI hand-over trick in recent Android networking
> stacks.
>